[OSRS-PROJ] Using PROJ4/LIBGEOTIFF with Lambert Conic Conformal 1SP projections on Algeria map data.
Willem Fourie (Businessware)
willem at businessware.cjb.net
Tue May 28 15:25:44 EDT 2002
I was required to use GeoTIFF 1:500 000 maps of areas in Algeria generated using Descartes/MicroStation, and to project them using PROJ4. The images have the following important GeoTIFF tags:
PCS = 30591 (Voirol Unifie / Nord Algerie)
Projection = 18021 (Nord Algerie)
Projection Method: CT_LambertConfConic_1SP
GCS: 4305/Voirol Unifie
Datum: 6305/Voirol Unifie 1960
Ellipsoid: 7012/Clarke 1880 (RGS) (6378249.14,6356514.87)
Prime Meridian: 8901/Greenwich (0.000000/000 00 00.00E)
Of course, looking through the file 'geotiff_proj4.c', it had to be unsupported. Undeterred, I implemented the GeoTIFF to PROJ4 conversion as follows, ala Gerald Evenden:
else if( psDefn->CTProjection == CT_LambertConfConic_1SP )
/* this appears to be an unsupported formulation with PROJ.4 */
/* to at least some degree this can be treated similarly to
* the 2SP case.
* See http://www.mentorsoftwareinc.com/CC/asknorm/ASK0699.HTM#Q2
/* Mods according to:
"+proj=lcc +lat_0=%.9f +lat_1=%.9f +lon_0=%.9f"
" +k_0=%.9f +x_0=%.3f +y_0=%.3f ",
This worked fine, except for what appeared like a scale creep effect when comparing my PROJ4 grid longitudes to those on the underlying map. Also, for the area I was testing, the PROJ4 latitudes were offset by a good 3 to 3.5 kilometers!
After some investigation it seems as if there is a bug in 'trf_nonpolynomial.c/csv' files. The value for k_0 is given as 0.999625544 for Nord Algerie and 0.999625769 for Sud Algerie. Other sources suggest these values should be 0.99625544 and 0.99625769, respectively.
I changed the values accordingly in the above files, and hey presto!, the underlying map grids and my PROJ4 generated grids overlapped pretty neatly.
Hope this is of some use to someone out there.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Proj